A LOAD-BALANCED ALGORITHM FOR MULTI-CONTROLLER PLACEMENT IN SOFTWARE-DEFINED NETWORK

摘要

Controllers in software-defined network (SDN) are amenable to flowchart generation and forwarding tasks on account of limited resource and connectivity. In fact, network partition and multicontroller placement will decide the network performance. Therefore, we proposed a novel multi-controller placement algorithm which can take into account load balance, connectivity and latency simultaneously. Firstly, the initial centroid node is selected not randomly but according to the minimized latency, which may lead to more robust clustering results. For load balance purpose, boundary nodes are allocated to adjacent cluster randomly. Then, isolated nodes are removed using tag delivery to keep connectivity. Via simulated annealing and iteration, the system gradually converges an optimized stable state, which is associating with several load-balanced sub-networks. Simulation results demonstrate that our approach significantly outperforms the state-of-the-art methods.